Advanced search
Start date
Betweenand
(Reference retrieved automatically from Web of Science through information on FAPESP grant and its corresponding number as mentioned in the publication by the authors.)

On strategies for building effective ensembles of relative clustering validity criteria

Full text
Author(s):
Jaskowiak, Pablo A. [1] ; Moulavi, Davoud [2] ; Furtado, Antonio C. S. [2] ; Campello, Ricardo J. G. B. [1] ; Zimek, Arthur [3] ; Sander, Joerg [2]
Total Authors: 6
Affiliation:
[1] Univ Sao Paulo, Dept Comp Sci, Sao Carlos, SP - Brazil
[2] Univ Alberta, Dept Comp Sci, Edmonton, AB - Canada
[3] Univ Munich, Database Syst & Data Min Grp, Munich - Germany
Total Affiliations: 3
Document type: Journal article
Source: KNOWLEDGE AND INFORMATION SYSTEMS; v. 47, n. 2, p. 329-354, MAY 2016.
Web of Science Citations: 8
Abstract

Evaluation and validation are essential tasks for achieving meaningful clustering results. Relative validity criteria are measures usually employed in practice to select and validate clustering solutions, as they enable the evaluation of single partitions and the comparison of partition pairs in relative terms based only on the data under analysis. There is a plethora of relative validity measures described in the clustering literature, thus making it difficult to choose an appropriate measure for a given application. One reason for such a variety is that no single measure can capture all different aspects of the clustering problem and, as such, each of them is prone to fail in particular application scenarios. In the present work, we take advantage of the diversity in relative validity measures from the clustering literature. Previous work showed that when randomly selecting different relative validity criteria for an ensemble (from an initial set of 28 different measures), one can expect with great certainty to only improve results over the worst criterion included in the ensemble. In this paper, we propose a method for selecting measures with minimum effectiveness and some degree of complementarity (from the same set of 28 measures) into ensembles, which show superior performance when compared to any single ensemble member (and not just the worst one) over a variety of different datasets. One can also expect greater stability in terms of evaluation over different datasets, even when considering different ensemble strategies. Our results are based on more than a thousand datasets, synthetic and real, from different sources. (AU)

FAPESP's process: 10/20032-6 - Study and development of validation methods for density- and graph-based clustering techniques
Grantee:Ricardo José Gabrielli Barreto Campello
Support type: Scholarships abroad - Research
FAPESP's process: 11/04247-5 - Gene selection and clustering validation in gene expression data
Grantee:Pablo Andretta Jaskowiak
Support type: Scholarships in Brazil - Doctorate
FAPESP's process: 12/15751-9 - Gene selection and Outliers in microarray data
Grantee:Pablo Andretta Jaskowiak
Support type: Scholarships abroad - Research Internship - Doctorate